Has anyone heard of an algorithm for managing keys automatically in a distributed system? For instance, if some low level security were to be implemented in a a networking stack where authentication was to be implemented, you would want to have each node have it's own signature so that signature checking can take place when one node connects to another node. The trick is then getting every node's keys distributed to every other node. Here are some ideas that I had, but neither is very desireable: - Manual distribution. User configures every node's key into every node. Configuration becomes a major hassle and mistakes are a pain to debug. An advantagous side effect is the user can configure which machines can talk to which machines if they're feeling particularly facist. - At connection time, each node determines whether or not it has the other node's key. If not, a symmetric key is generated via DH and public keys are exchanged. The problem comes in if someone is spoofing the machine to begin with. Then you'll have the wrong public key. Chicken, egg. Egg, chicken. - Having a certifying node which every other node has the public key to and who has everyone else's public key. Requests are made of this server. The trick is making this server secure and forcing the user to devote resources to this endeavour. Thoughts? -- Matt Smith - msmith@unislc.slc.unisys.com "Nothing travels faster than light, with the possible exception of bad news, which follows its own rules." - Douglas Adams, "Mostly Harmless" Disclaimer: I came up with these ideas, so they're MINE!
Since no one seemed to mention them, check out Photuris and SKIP. Both are key management proposals for IPsec. Adam Matt Smith wrote: | | | Has anyone heard of an algorithm for managing keys automatically in a | distributed system? | | For instance, if some low level security were to be implemented in a | a networking stack where authentication was to be implemented, you would want | to have each node have it's own signature so that signature checking can | take place when one node connects to another node. The trick is then | getting every node's keys distributed to every other node. | | Here are some ideas that I had, but neither is very desireable: | | - Manual distribution. User configures every node's key into every node. | Configuration becomes a major hassle and mistakes are a pain to debug. | An advantagous side effect is the user can configure which machines can | talk to which machines if they're feeling particularly facist. | | - At connection time, each node determines whether or not it has the other | node's key. If not, a symmetric key is generated via DH and public keys | are exchanged. The problem comes in if someone is spoofing the machine | to begin with. Then you'll have the wrong public key. Chicken, egg. | Egg, chicken. | | - Having a certifying node which every other node has the public key to and | who has everyone else's public key. Requests are made of this server. | The trick is making this server secure and forcing the user to devote | resources to this endeavour. | | Thoughts? | | -- | Matt Smith - msmith@unislc.slc.unisys.com | "Nothing travels faster than light, with the possible exception of bad news, | which follows its own rules." - Douglas Adams, "Mostly Harmless" | Disclaimer: I came up with these ideas, so they're MINE! | -- "It is seldom that liberty of any kind is lost all at once." -Hume
On Wed, 15 May 1996, Matt Smith wrote:
- Having a certifying node which every other node has the public key to and who has everyone else's public key. Requests are made of this server. The trick is making this server secure and forcing the user to devote resources to this endeavour.
This is the usual approach; if you use certificates, the private key for the certification agency doesn't have to be (and shouldn't) be accessible online; thus, even if the machine serving the certificates is compromised, Mallet won't be able to issue false certificates. --- Cause maybe (maybe) | In my mind I'm going to Carolina you're gonna be the one that saves me | - back in Chapel Hill May 16th. And after all | Email address remains unchanged You're my firewall - | ........First in Usenet.........
participants (3)
-
Adam Shostack -
Matt Smith -
Simon Spero